he undercover 
a result of 
ty or total 
that increased 
s from heavy 
pectrum are 
ole of vegeta- 
ne type develo- 
st be used. The 
age and are 
afore, 
permit rapid 
it 20m on a 
for the Chiu 
ton (1976), 
1s and Chiu 
allins multi- 
a and vegeta- 
report increased 
^ lodgepole pine 
posit. While 
the chlorophyll 
author and 
Howeyer, these 
1e spectral 
decrease in 
je and 
reflectance 
> nm) and pro- 
ip plant species 
je of both 
colors corres- 
Ihe areas 
yrrelated with 
1 aerial 
| area of Cu 
scence outside 
lone to date, 
ig for detec- 
jlant density 
  
iy 
AAN j 
” > 
  
  
LANDSAT STUDIES 
The Landsat satellite lacks the high spatial and spectral resolution of the land 
based and airborne studies ; nevertheless, it is possible to detect geobotanical 
anomalies from space. Lyon (1975) used a number of Landsat band ratios, of which. 
B7/B4 was best, to discriminate a juniper and pine geobotanical anomaly associa- 
ted with a skarn molybdenum deposit. Bolviken et al. (1977) detected stands of 
Cu stressed vegetation on enhanced color ratio images of Landsat digital data : 
and Lefevre. (1982) used Landsat B7/B6 and B7/B5 ratios to detect low vegetation 
that correlated with geochemical anomalies of As. 
  
This author and his coworkers (Stone, 1982 ; Stone et al., 1982) have used 
Landsat digital data to map potential mineralized zones to the southeast of the 
Caraiba Mining District, Bahia State Brazil. The Cu-Ni mineralization in this 
district is associated with mafic and ultramafic rocks intruded into Precambrian 
metasediments. The ultramafic units weather in this semi-arid environment into 
dark brown clay rich soils that contrast significantly with the light toned 
quartz rich soils developed onthe surrounding metamorphic rocks. Metamorphic 
terranes are particularly difficult to map using remote sensing techniques 
because the lithologic boundaries are more gradational and less distinct than 
other terranes (Abrams, 1980). Geobotanical techniques are particularly useful 
if they produce distinct boundaries. Low, bushy and succulent plants predominate 
the Brasilian study area ; and Lewis (1966) noted a strong geobotanical prefe- 
rence of certain plants to either the clay rich soils derived from the mafic 
rocks or the sandy soils of the metamorphic rocks. The total biomass growing 
on the mafic soils is also greater (Putzer, 1976). 
  
Numerous investigators have applied various vegetation indices for the purpose 
of biomass determination (for example Tucker, 1979) ; these indeces exploit the 
fact that solar radiance in the visible red band (B5) is strongly absorbed by 
chlorophyll, whereas the near infrared. radiance is strongly reflected. After 
testing a number of vegetation indices, it was determined that the indices that 
merely examine the difference between the near infrared and visible bands (for 
example, the DVI of Richardson and Wiegand, 1977 where DVI = 2. 4B7-B5 (Fíg.,2) 
did not accurately reflect the fact that the mafic rock units were more densely 
vegetated. Since in this semi-arid region the vegetation cover is generally not 
complete,there is a considerable amount of soil integrated into each pixel ; 
and the soils derived from the felsic rocks are brighter than those of the mafic 
rocks. The data presented by Rowan et al. (1974) (Fig. 3) indicate that felsic 
rocks have greater B7-B5 values. The low vegetation density over the gneissic 
units allows more soil to show through, and the soils derived from these felsic 
rocks have high "apparent" vegetation indices relative to the mafic derived 
soils. 
When a standard soil brightness index is applied to the data (for example SBI 
of Dauth and Thomas (1976) where SBI = 0.433B4+0.632B5+0.506B6+0.264B7 (Fig. 4) 
the darkness of the mafic soils is clearly manifest. Therefore, when different 
soils occur under a vegetation canopy, a simple difference biomass indicator 
must be normalized to account for soil brightness. The vegetation index of 
Tucker (1979) (VI =(B7-B5)/(B7+B5)) does this and results in an index that shows 
a strong correlation of increased biomass and mafic rocks (Fig. 5). 
The overall index can be improved in this case where dark soils occur on 
units with high vegetation density by ratioing the vegetation and soil brightness 
indices (VI/SBI (Fig. 6) ). The inverse correlation of the soil brightness and 
vegetation indices produces a striking separation of mafic and gneissic units. 
In many remote sensing studies, naturally occurring vegetation may be a problem 
because it masks the spectral response of the earth materials (Siegal and Goetz, 
169 
I ns SCS ST o s d